Define the long-term security architecture and technical patterns for autonomous agents, MCP integrations, multi-turn reasoning, and multi-agent coordination.Architect and deploy runtime guardrails, semantic firewalls, intent verification filters, and policy-enforcement systems.Design secure, short-lived, micro-isolated execution environments for agent code, shell commands, and tool